Instrumental Communities and the …users.nber.org/~sewp/ModyInstruments2-06.doc · Web...
Transcript of Instrumental Communities and the …users.nber.org/~sewp/ModyInstruments2-06.doc · Web...
Instruments of Commerce and Knowledge: Probe Microscopy, 1980-2000
Cyrus C. M. Mody
Chemical Heritage Foundation
Introduction
The voices of editorialists and analysts excoriating or praising commercialization
of products of higher education has recently grown very loud.1 Yet this debate too often
proceeds at an abstract level divorced from the small-scale settings where
commercialization actually occurs. The questions asked are too stark, and the dangers
and benefits of academic entrepreneurialism are amplified beyond recognition. As
Steven Shapin notes in a recent survey, proponents’ and nay-sayers’ limited historical
horizons lead to, among other things, ludicrous over-praising of incentives to patent
academic research and over-dire warnings about the corporate university.2 The few
historical studies that could contribute to this debate, while praiseworthy, have
My thanks to Mike Lynch, Arthur Daemmrich, Steve Shapin, John Staudenmaier, and David Kaiser for their advice and encouragement on various drafts of this paper. Audiences at Arizona State, the American Sociological Association, the National Bureau of Economic Research, and the Chemical Heritage Foundation also provided useful comments. Portions of this work have appeared in Technology and Culture. This work was made possible by funding from NBER, the NSF, and the IEEE History Center, as well as by the generous cooperation of my interviewees.1 See Norman E. Bowie, ed., University-Business Partnerships: An Assessment, Issues in Academic Ethics (Lanham, MD: Rowman & Littlefield, 1994); Derek Bok, Universities in the Marketplace: The Commercialization of Higher Education (Princeton: Princeton University Press, 2003); Roger L. Geiger, Knowledge and Money: Research Universities and the Paradox of the Marketplace (Stanford: Stanford University Press, 2004); Frank Newman, Lara Couturier, and Jamie Scurry, The Future of Higher Education: Rhetoric, Reality, and the Risks of the Market (San Francisco: Jossey-Bass, 2004); David L. Kirp, Shakespeare, Einstein, and the Bottom Line: The Marketing of Higher Education (Cambridge, MA: Harvard University Press, 2003); Eric Gould, The University in a Corporate Culture (New Haven: Yale University Press, 2003); and the essays in Donald G. Stein, ed., Buying In Or Selling Out: The Commercialization of the American Research University (New Brunswick: Rutgers University Press, 2004) and Joseph C. Burke, ed., Achieving Accountability in Higher Education: Balancing Public, Academic, and Market Demands (San Francisco: Jossey-Bass, 2004).2 Steven Shapin, “Ivory Trade,” London Review of Books 25, no. 17 (2003): 15-19.
1
concentrated too narrowly on a handful of particularly entrepreneurial universities
(Stanford and MIT), disciplines/industries (microelectronics and biotechnology), and
regions (Silicon Valley and Route 128 near Boston).3
Yet these studies neglect important aspects of participants’ experience of
corporate-academic cooperation. Most researchers participate in networks that are
geographically dispersed and that include colleagues in both academia and industry and
from a variety of disciplines. To understand the commercialization of academic
knowledge, we need a multi-institutional, multi-disciplinary, multi-regional unit of
analysis – what I will call an “instrumental community.” By this I mean the porous
group of people commonly oriented to building, developing, using, selling, and
popularizing a particular technology of measurement.4 Such communities are
“instrumental” primarily in focusing on new research tools – microscopes, fruit flies,
tobacco mosaic virus, lab rats, cathode ray tubes, etc.5 Because such communities
3 For MIT and Stanford, see Stuart W. Leslie, The Cold War and American Science: The Military-Industrial-Academic Complex at MIT and Stanford, 1 ed. (New York, NY: Columbia University Press, 1993); John Servos, “The industrial relations of science: Chemical engineering at MIT, 1900-1939,” Isis 81 (1980): 531-49; R. S. Lowen, “Transforming the University - Administrators, Physicists, and Industrial and Federal Patronage At Stanford, 1935-49,” History of Education Quarterly 31, no. 3 (1991): 365-388; and C. Lecuyer, “Academic science and technology in the service of industry: MIT creates a "permeable" engineering school,” American Economic Review 88, no. 2 (1998): 28-33. For biotech and microelectronics, see Martin Kenney, Biotechnology: The University-Industrial Complex (New Haven: Yale University Press, 1986) and Christophe Lécuyer, Making Silicon Valley: Innovation and the Growth of High Tech, 1930-1970 (Cambridge, Mass., 2006). For a regional perspective, see Anna-Lee Saxenian, Regional Networks: Industrial Adaptation in Silicon Valley and Route 128 (Cambridge, MA: Harvard University Press, 1993); Peter Hall and Ann Markusen, eds., Silicon Landscapes (Boston: Allen & Unwin, 1985).4 The “instrumental community” bears a close resemblance to the “innovation communities” analyzed by Sonali K. Shah, “Open Beyond Software,” in Open Sources 2.0: The Continuing Evolution, ed. Danese Cooper, Chris DiBona, and Mark Stone (Sebastopol, CA: O’Reilly Media, 2005). “Instrumental community” is – so far as I know – my own formulation, but others have covered very similar ground, especially: Stuart Blume, Insight and Industry: On the Dynamics of Technological Change in Medicine (Cambridge, MA: MIT Press, 1992) and Terry Shinn, “Crossing Boundaries: The Emergence of Research-Technology Communities,” in Universities and the Global Knowledge Economy: A Triple Helix of University-Industry-Government Relations, ed. Henry Etzkowitz and Loet Leydesdorff (London: Pinter, 1997), 85-96.5 For studies in this vein, see: Robert Kohler, Lords of the Fly (Chicago: University of Chicago Press, 1994), Boelie Elzen, “Two Ultracentrifuges: A Comparative Study of the Social Construction of Artefacts,” Social Studies of Science 16 (1986): 621-662; Karen Rader, Making Mice: Standardizing
2
usually include academic and commercial participants, though, they will often seek ways
to morph those tools into industrially-relevant devices. Thus, such communities are also
“instrumental” in focusing on new ways of doing or making things.
There are a number of excellent case studies of various instrumental communities,
spanning from the seventeenth century to the 1960s.6 Yet there have been virtually no
studies of instrumental communities that have arisen since the late 1970s. This is
unfortunate in that this is the period on which the most overheated rhetoric about
academic capitalism is focused. We know that there have been significant changes in
legislation, federal funding, corporate research, and the demographics of science in the
past three decades.7 We do not know how those changes have affected the operation of
instrumental communities, nor how they have affected relationships between corporate
and academic members of those communities. This chapter aims to bring these issues to
the fore through a case study of the development and commercialization of the scanning
tunneling microscope (STM) and its near-relatives, the atomic force microscope (AFM)
and magnetic force microscope (MFM) – known collectively as probe microscopes.8
Animals for American Biomedical Research, 1900-1955 (Princeton: Princeton University Press, 2004).6 See, chronologically, Steven Shapin and Simon Schaffer, Leviathan and the Air-Pump: Hobbes, Boyle, and the Experimental Life (Princeton: Princeton University Press, 1985); Myles W. Jackson, “Buying the dark lines of the spectrum: Joseph von Fraunhofer’s standard for the manufacture of optical glass,” in Scientific Credibility and Technical Standards in 19th and Early 20th Century Germany and Britain, Jed Z. Buchwald, ed. (Dordrecht: Kluwer Academic, 1996), 1-22; and David Pantalony, “Seeing a voice: Rudolph Koenig's instruments for studying vowel sounds,” American Journal of Psychology 117, no. 3 (2004): 425-442; Nicolas Rasmussen, Picture Control: The Electron Microscope and the Transformation of Biology in America, 1940-1960 (Stanford: Stanford University Press, 1997); Joan Lisa Bromberg, The Laser in America, 1950-1970 (Cambridge, MA: Massachusetts Institute of Technology Press, 1991); and Timothy Lenoir and Christoph Lécuyer, “Instrument Makers and Discipline Builders: The Case of Nuclear Magnetic Resonance,” Perspectives on Science, no. 3 (1995): 276-345.7 As outlined in Philip Mirowski and Esther-Miriam Sent, “STS, the Economics of Science, and the Commercialization of University Science,” ed. Edward Hackett, Olga Amsterdamska, Michael Lynch, and Judy Wacjman (Cambridge, MA: MIT Press, forthcoming).8 The technical details of the microscopes are important to this story, but can be glossed for the purposes of this chapter. Basically, all scanning probe microscopes bring a very small solid probe very close (usually to within a nanometer – one billionth of a meter) to a sample and measure the strength of different kinds of interactions between probe and sample to determine the height (and other characteristics) of the sample. The probe is then rastered much like the pixels on a TV screen and a matrix of values for the strength of the
3
Twenty-five years ago there was one, home-made, unreliable STM at the IBM research
lab in Zurich. Today, through the joint efforts of corporate and academic researchers,
there are thousands of AFMs, MFMs, and STMs at universities, national labs, and
industrial research and quality control facilities. High school students make STMs from
Legos, while chip manufacturers use million-dollar AFMs on the factory floor. One
AFM has even made it to the surface of Mars.
Inventing and Community-Building
Invention can be a precarious business, particularly for corporate scientists and
engineers.9 Inventions often emerge from digressions from assigned tasks, and may not
initially meet any commercial objective. The STM, for one, was this kind of institutional
orphan. Its inventors, Gerd Binnig and Heini Rohrer, had been tasked with finding new
ways to characterize thin films used in an advanced supercomputer project on which IBM
had staked much of its reputation. Yet by the time they came up with the STM as an
tip-sample interaction is converted into a visual “picture” of the surface. Different probe microscopes use different kinds of tip-sample interactions to generate their images. The first, the STM, works by putting a voltage difference between the tip and a metal or semiconductor sample; when the tip is brought close to the sample, some electrons will quantum mechanically “tunnel” between them. The number of electrons that do so (the “tunnel current”) is exponentially dependent on the distance between tip and sample; also, the stream of tunneling electrons is very narrow. Thus, an STM has ultrahigh resolution both vertically and laterally – most STMs can actually see individual atoms on many samples. Today, the STM’s younger cousin, the atomic force microscope, is more commonly used. An AFM uses a very small but flexible cantilever as a probe; as the tip of the cantilever (usually weighted with a small pyramid of extra atoms) is brought close to the surface, the cantilever bends due to the attraction or repulsion of interatomic forces between tip and sample. The degree of bending is then a proxy for the height of the surface. Originally this bending was measured by putting an STM on the back of the cantilever; today the deflection is detected by bouncing a laser off the cantilever and measuring the movement of the reflected spot. Another common and industrially-relevant tool, the magnetic force microscope, works in a similar way, but uses a magnetic tip to map the strength of magnetic domains on a surface, rather than surface height. Both the AFM and MFM have slightly less resolution than the STM (i.e. they cannot usually see single atoms); yet because they (unlike the STM) can be used on insulators as well as conductors, and in air and fluids as well as vacuum, they have become much more popular.9 Indeed, inventors of instruments (or those who take credit for having invented them) often seem to have vexed positions within the firms that employ them. See, for instance, the description of Kary Mullis’ antagonistic relationship with Cetus in Paul Rabinow, Making PCR: A Story of Biotechnology (Chicago: University of Chicago Press, 1996).
4
answer to the thin film problem, the supercomputer project had been canceled.10 Binnig
and Rohrer’s response was three-fold. First, they hid the STM from managerial view –
easy enough at the Zurich lab, far from corporate headquarters and well-known for lax
oversight. Second, they began querying IBM colleagues about new ways to use the
STM, eventually attracting interest from the company’s large cadre of semiconductor
surface scientists.
Their third, key strategy was to cultivate an extramural, academic community
committed to the STM by encouraging their network of acquaintances to replicate the
instrument. As this instrumental community grew both inside and outside Big Blue,
IBM’s senior research managers decided that the STM – despite the absence of
commercial relevance – should become a major corporate project. Multiple groups of
scientists at the IBM laboratories in Zurich, Yorktown Heights, New York and Almaden,
California were recruited out of graduate school to build STMs and make discoveries that
would bring credit to the instrument and to the company. In turn, IBM’s research
archrival, Bell Labs, saw a need to steal Big Blue’s thunder and began recruiting its own
cadre of STMers.
The dynamics of building the STM community show how the corporate and
academic worlds are interpermeated much more thoroughly and enduringly than is often
noticed in debates about academic commercialization. Binnig and Rohrer could quickly
cultivate a set of academic STM replicators because of networks of personnel exchange
between IBM and various universities – some replicators were professors taking
sabbaticals in Zurich, some were academics Rohrer had known from his own sabbaticals
10 G. Binnig and H. Rohrer, “The Scanning Tunneling Microscope,” Scientific American 253, no. 2 (1985): 50-6 and G. Binnig and H. Rohrer, “Scanning Tunneling Microscopy - From Birth to Adolescence,” Reviews of Modern Physics 59, no. 3 (1987): 615-625.
5
at universities, and some were people who had been postdocs at IBM or currently had
students serving postdoctoral appointments there.11
Similarly, interest in the STM grew within IBM and Bell Labs not because it
could solve commercially-relevant problems, but because it could generate credible
knowledge within academic disciplines such as physics and surface science.12 Accolades
from an academic audience – evidenced by standing-room-only crowds at American
Physical Society meetings, awarding of the Nobel Prize to Binnig and Rohrer in 1986,
and the growth of academic STM – were largely the aim of IBM’s STM program.
Moreover, prestige within a hot, new instrumental community like STM in turn allowed
IBM to recruit the best graduate students as postdocs and junior researchers – exactly the
people who built the second and third generations of IBM’s tunneling microscopes.
Today, some of those same people have returned IBM’s investment by becoming the
leading figures in nanotechnology research – securing Big Blue’s reputation and
intellectual property in what is, at last, a commercially important area.
Dynamics of Community
By 1986, then, the STM was no longer a precarious technology. Crucially, the
instrumental community beginning to take shape was dominated by corporate groups,
11 The source material for this study is a collection of interviews with 150+ probe microscopists conducted between 2000 and 2004. I will reference specific oral histories using an alphanumeric code listed in the appendix to this article. Information about the corporate-academic network of sabbaticals and hires came from, among others, <TB1>, <JM1>, and <PH1>.12 There is rich historical material on the large, corporate labs of the twentieth century: George Wise, Willis R. Whitney, General Electric, and the Origins of U.S. Industrial Research (New York: Columbia University Press, 1985); Michael Riordan and Lillian Hoddeson, Crystal Fire: The Birth of the Information Age (New York: Norton, 1997); Lillian Hartmann Hoddeson, “The roots of solid-state research at Bell Labs,” Physics Today (1977); Leonard Reich, The Making of American Industrial Research: Science and Business at GE and Bell, 1876-1926 (Cambridge, UK: Cambridge University Press, 1985). Most relevant here are analyses of the tenuous relationship between research and production at IBM: Ross Knox Bassett, To the Digital Age: Research Labs, Start-Up Companies, and the Rise of MOS Technology (Baltimore: Johns Hopkins, 2002); Scott Knowles and Stuart W. Leslie, “’Industrial Versailles’ – Eero Saarinen’s Corporate Campuses for GM, IBM, and AT&T,” Isis 92: 1-33.
6
especially from IBM and Bell Labs. Early academic STMers, such as Paul Hansma at the
University of California at Santa Barbara (UCSB), Calvin Quate at Stanford, and John
Baldeschwieler at Caltech, were important contributors to the community; yet these
academics struggled to compete with corporate groups that were better-resourced and
(more importantly) were working alongside (or in competition with) numerous other
STMers in the same building. The tacit knowledge needed to build an STM flowed more
quickly at IBM and Bell Labs, allowing those organizations to rapidly expand their
commitment to STM.13
This meant the questions most important to the early STM community were those
of relevance to groups at IBM and Bell Labs. In particular, since Binnig and Rohrer had
been most successful in enrolling colleagues interested in the surface structure of metals
and semiconductors, those the community’s chosen materials. Indeed, a few surfaces
(especially of silicon) served as yardsticks for measuring whether a group had a working
STM or not – until a group’s STM had resolved single atoms of silicon, its builders could
not enter the top tier of STM builders.14 Other metal and semiconductor surfaces served
as milestones, with different groups racing each other to be the first to achieve atomic
resolution. Thus, interest in semiconductors – obviously strong at IBM and Bell Labs –
helped standardize activity in the community and allowed participants to judge each
other’s progress.
Initially, academics such as Quate and Baldeschwieler tried to keep up in these
races. Notably, Quate, located near both Silicon Valley and a cadre of former students
13 For treatments of the concept of tacit knowledge, especially as applied to instrument-building, see H. M. Collins, “The Seven Sexes: A Study in the Sociology of a Phenomenon, or the Replication of Experiments in Physics,” Sociology 9, no. 2 (May) (1975): 205-224; Harry Collins, “Tacit Knowledge, Trust, and the Q of Sapphire,” Social Studies of Science 31, no. 1 (2001): 71-86; and Michael Polanyi, Personal Knowledge: Towards a Post-Critical Philosophy (New York: Harper Torchbooks, 1962).14 <JD2>, <PW2>.
7
and postdocs at IBM Almaden, had the most success in this area. Yet even he struggled
to develop familiarity in handling metals and semiconductors among his students and in
making his results credible to corporate STMers.15 Hansma, meanwhile, saw that IBM
and Bell Labs would continue to dominate the study of metal and semiconductor surfaces
and began carving alternative niches.16 Soon, Quate, Baldeschwieler, and other
academics followed suit, so that the STM community began to segregate into two
moieties – surface science STMers, dominated by (but not exclusive to) corporate and
national laboratories on the East Coast; and non-surface scientists, dominated by (but not
exclusive to) universities on the West Coast.17
These two moieties continued to share a great deal. Members of each
occasionally collaborated, and a few people moved from one to the other. More
importantly, the basic design of the STM was – at this stage – common to both, so design
innovations in one moiety could be transported to the other. This meant that
opportunities for co-presence – conferences and visits and sabbaticals between labs –
continued to be useful. Yet in a number of areas the two moieties established starkly
different styles. In particular, corporate STMers, coming out of a surface science
tradition that valued ultraclean samples and rigid control of conditions, built their STMs
for compatibility with ultrahigh vacuum (UHV) chambers. These chambers were large,
finicky, expensive, and time-consuming, so academic STMers developed alternatives
such as STM in air, in water, in oil, and in a variety of gases.18
15 <JD2>.16 <PH1>, <BD2>.17 Crucial members of the latter, predominantly academic, moiety were Quate’s allies within IBM: Dan Rugar, John Foster, and Tom Albrecht (former students who worked at IBM Almaden); Kumar Wickramasinghe (a former postdoc, later at IBM Yorktown); and Gerd Binnig (who took a sabbatical at Stanford in 1985-6).18 <CP1>, <PH1>.
8
Freed from the constraints of UHV and the need to please surface scientists,
academic STMers moved to a radically open-ended, sometimes chaotic mode of
experimentation. They saw that corporate surface science STM had succeeded by
appealing to specific disciplinary audiences and by orienting to a few yardstick materials
by which members of the community could be measured, and sought out new audiences
and yardstick materials of their own. It was unclear, though, which audiences might
accept the STM, and how the STM should be adapted to achieve acceptance. Thus,
academics like Quate and Hansma encouraged their students to quickly build a wide
variety of microscopes and to playfully use them to characterize haphazard materials –
leaves of houseplants, polaroids, bone from ribeye steaks, ice, the electrochemistry of
Coke versus Pepsi, etc.19 This bricolage fit well with these groups’ shoestring operation
(in contrast to the corporate groups) and extended even into microscope-building: the
Baldeschwieler group made STM probes from pencil leads, for instance, while the
Hansma group made AFM tips from hand-crushed pawn shop diamonds, glued to tin foil
cantilevers with brushes made from their own eyebrow hairs.
Yet such indiscipline could damage STM’s acceptance by new disciplinary
audiences, since the microscope-builders did not know how to prepare samples and
interpret images in ways that would be credible to, for example, biologists,
electrochemists, or materials scientists. Thus, Quate, Hansma, and other academic
STMers began bringing representatives (postdocs or young professors) from potential
new disciplinary audiences in to work with their students, learn how to use the
microscope, show the group how to prepare samples, and then proselytize for the
technique within their home community. Often, these people took a microscope with
19 <CP1>, <JN1>.
9
them when they left, or founded their own microscope-building group, and used their
knowledge of probe microscopy as a tool for gaining prestige among disciplinary
colleagues and securing tenure from their universities.20
Thus, the differences between the two moieties were as much about pedagogy and
career arc as they were about samples, designs, and audience. In groups such as Quate’s
and Hansma’s (mirrored, in part, by their collaborators’ groups), graduate students were
trained to build instruments quickly and collaboratively, to think primarily about novel
design rather than use; postdocs, meanwhile, were trained to develop new uses out of
those designs, and to integrate a new technology into an established discipline – STM for
biology or materials science or electrochemistry. In the corporate labs, postdocs, too,
underwent a kind of training – in a position of constant oversight by colleagues and
managers with the expertise and power to judge their work and affect their careers,
corporate postdocs learned to build and use STMs geared specifically to institutional
needs. Thus, corporate surface science STMs all looked relatively similar and were used
to look at the same handful of samples – though with enough variation to demonstrate
their builders’ personal qualities of initiative, creativity, and experimental ingenuity.21
In other words, the instrumental community growing around the STM included
elements of pedagogy at all participating sites, rather than just in the academic groups –
the STM was a technology for turning young researchers into full-fledged scientists as
much as a new technique for characterizing materials. Analysts of academic capitalism
should keep this in mind – universities have no monopoly on training in science, and the
20 <AG1>, <HG1>, <JN1>. The propagation of a technique through the “cascade” of postdocs and collaborators away from one of the centers of an instrumental community is described in David Kaiser, Drawing Things Apart: The Dispersion of Feynman Diagrams in Postwar Physics (Chicago: University of Chicago Press, 2005).21 <BW2>.
10
paths of pedagogy often seamlessly cross from the academic to the corporate world.
Moreover, in this particular case the pedagogical uses of the STM encouraged a wider
division of labor in the instrumental community. Because corporate postdocs were
promoted based on their ability to integrate the STM with surface science, graduate
students building STMs were instead encouraged to expand the instrument’s capabilities
into new areas. Corporations did, indeed, “influence” the direction of academic research
here; but whereas many critics decry the constrictive, harmful effects of corporate
influence, in this case corporate actors encouraged academics to adopt a diversity of
approaches and an expansive outlook.
Building and Buying
In this sense, pedagogy drove the two moieties of early STM apart; yet in a
variety of ways, corporate and academic participants were inseparable, particularly
around practices of building microscopes and finding samples to characterize. Until
1986, all probe microscopes were “home-built” in that they were put together by the
groups that were using them. Yet home-built instruments were not made entirely from
scratch – some components were made by hand, but most were bought from commercial
suppliers. STM designs were strongly shaped by the commercial availability of
components such as op amps and probe materials; but STM builders were also active
consumers. They took commercial products and adapted them for unforeseen uses and
negotiated with suppliers for equipment (vacuum chambers, piezoelectric crystals, video
output devices, etc.) modified for their specific applications.22 Some suppliers, such as
22 Much recent history of technology has focused on the active role of users. For consumers’ adaptations of artifacts for uses that manufacturers were unaware of, or even opposed, see Ronald Kline and Trevor Pinch, “Users as Agents of Technological Change: The Social Construction of the Automobile in the Rural United States,” Technology and Culture 37, no. October (1996): 763-95. For users’ pressure on companies (often – as in instrumental communities – through threats to form their own cooperatives or firms), see Claude S. Fischer, America Calling: A Social History of the Telephone to 1940 (Berkeley: University of California
11
Burleigh (a piezoceramic maker in upstate New York) modified their products, using the
advice of prominent STMers, so that they could sell the modified items specifically for
the STM market.23 That is, professors were crucial both as consumers and producers.
Most favorable analyses of academic capitalism have focused on academics as producers,
with a view to stimulating professorial start-up companies; yet universities may find that
the best way to gain a stake in an instrumental community is by encouraging professors
to be smart, savvy, active consumers who have leverage with manufacturers both through
their expertise and their potential for building their own product (or even starting a
competing firm).
In a number of indirect and often counter-intuitive ways, commerce supplied the
infrastructure of knowledge and standardization needed to make the STM community
grow. Information about sources of reliable components and materials was a major topic
of “gossip” among early STM builders. Those who had built working instruments
recommended particular brands to new members of the community; and those
newcomers, anxious to make up for lost time, rarely questioned their predecessors’
advice. Old-timers gladly gave newcomers blueprints and recommendations on
suppliers, so that STM-building came to resemble doing a project from Popular
Mechanics. Brands became an important carrier of the tacit knowledge of microscope-
building, and STM-building became standardized through STMers’ nearly ritualistic
allegiance to those brands, even when the technical rationale for the brand disappeared.
For instance, IBM’s STMers used a trademarked rubber called Viton (from Dupont) to
Press, 1992). For an overview of different kinds of user activity, see the essays in Nelly Oudshoorn and Trevor Pinch, eds., How Users Matter: The Co-Construction of Users and Technologies (Cambridge, MA: MIT Press, 2003).23 <DF1>.
12
dampen vibration, because Viton could survive ultrahigh vacuum.24 Later, as IBM’s
blueprints disseminated, Viton became a hallmark of tunneling microscopy, even in
academic STMs used in air or fluid, not vacuum.25
A commercial infrastructure also helped STMers standardize the materials they
characterized. As Daniel Lee Kleinman has noted, this kind of corporate “influence” on
research is pervasive but indirect.26 Yet tapping into the right commercial infrastructure
can be crucial to growing an instrumental community and ensuring the credibility of its
members’ expertise – reliable, cheap commercial sources of materials give newcomers
easy access to research, and give the rest of the community a yardstick by which to
measure newcomers’ progress. Among corporate surface science STMers, this yardstick
was provided by a few key metal and semiconductor samples on which newcomers had to
prove their machines. When academic STMers designed microscopes for use in air and
water, they needed alternative yardstick materials. Gold, paraffin, and graphite vied for
the job; but graphite won out partly because ultrapure samples could be obtained
cheaply.27 Union Carbide used graphite to make monochromators for neutrons, an
application requiring extraordinarily pure samples – hence, they rejected large amounts of
slightly imperfect graphite still pure enough for STMers. The Quate group heard about
this and alerted other academic groups who then called Union Carbide’s graphite man,
Arthur Moore, to get cheap, standardized samples. Soon, the STM community was
24 <CG1>, <RT1>, <VE1>.25 For similar instances of practices spreading through an experimental community through transmission of knowledge about particular brands, see Kathleen Jordan and Michael Lynch, “The dissemination, standardization, and routinization of a molecular biological technique,” Social Studies of Science 28, no. 5-6 (1998): 773-800.26 Daniel Lee Kleinman, Impure Cultures: University Biology and the World of Commerce (Madison: University of Wisconsin Press, 2003).27 <AG1>.
13
awash in graphite, such that talks about that material finally outnumbered talks on
semiconductors at the annual STM conferences.
Sometimes, the industrial relevance of materials was a more direct influence
acting on academic microscopists, feeding back into the designs of their instruments. Yet
the arrow of corporate-academic influence here was dramatically non-linear. Cal Quate,
for instance, framed his STM work within Stanford’s long tradition of industrial ties and
his own involvement in developing acoustic microscopy in the ’70s as a non-destructive
characterization tool for manufacturing.28 Non-destructive testing held tremendous
promise for microelectronics, where chips are inspected throughout manufacturing yet
where traditional tools (especially electron microscopy) require breaking and discarding
expensive silicon wafers. Quate moved into STM believing it could be the next
generation non-destructive evaluation tool; and he was quickly followed by his former
students and postdocs at IBM.29
STM, though, requires a conducting (metal or semiconductor) sample, whereas
most microelectronic materials have an insulating oxide layer. Indeed, controlled growth
of oxides is crucial to turning silicon wafers into integrated circuits. This was
unproblematic for corporate surface scientists tasked with generating basic knowledge
about materials like silicon and gallium arsenide. Yet STM’s restriction to conducting
materials blocked its use in non-destructive testing and hindered movement into fields
other than surface science. Those who wanted to carve interdisciplinary niches for STM
saw its sample constraints as suffocating; chief among these were Gerd Binnig (an IBM
28 <JF1>, <DR1>, <MK1>. See C. F. Quate, “Acoustic Microscopy - Recollections,” IEEE Transactions On Sonics and Ultrasonics 32, no. 2 (1985): 132-135 for a brief description of scanning acoustic microscopy at Stanford.29 Quate’s optimism for STM derived from its ultrahigh resolution and the fact that (ideally) the STM tip does not touch (and thereby mar) the sample surface.
14
employee but not a surface scientist) and Cal Quate (and his former students and postdocs
at IBM). So when IBM allowed Binnig to take a sabbatical at Stanford in 1985-6, he and
Quate pushed past the STM to invent the AFM – which, because it uses interatomic
forces rather than tunneling to sense height, can map insulating materials. Thus, Quate
positioned his research much further downstream in IBM’s manufacturing cycle than
most of IBM’s own STMers and, together, IBM and Stanford dramatically shifted the
world of academic and corporate probe microscopy.
Commercialization and Gray Markets
What we’ve seen so far, then, are the more intricate, unglamorous ways corporate
and academic actors are linked within an instrumental community – through pedagogy,
through institutional politics, through commercial infrastructures, through tacit
knowledge. Most of relationships I’ve outlined thus far are not the ones that exercise
proponents and critics of academic capitalism – large corporate buy-ins to academic
departments, professors keeping research secret so they can patent it, corporations and
universities colluding to suppress unfavorable results.
One topic central to the academic capitalism debate will occupy the rest of this
paper – the commercialization of academic research and the founding of professorial
start-up companies. Yet I will show that commercialization only happens when it helps
participants position themselves within an instrumental community; and that the
commercialization process is not sudden and dramatic, but built slowly and quietly from
the kinds of practices I have described thus far.
The proximate basis for commercialization of probe microscopy was the desire on
the part of elite STMers to grow a larger community in which their groups would be
15
centers of expertise.30 This desire was strong among both corporate and academic
groups, though the motivations differed in the two moieties. At IBM and (to a lesser
extent) Bell Labs, research managers wanted to increase the number of in-house STM
groups, so as to keep the center of the STM community within the corporation.
Academics like Quate and Hansma wanted to grow the STM (and AFM) community
because they were looking for new applications and audiences, and because they wanted
to build a critical mass committed to non-surface science probe microscopy.
Thus, the stimulus to (a kind of) commercialization existed in both the corporate
and academic environments, as did the practices and knowledge from which this proto-
commercialization could be constructed. Both Bell Labs and IBM, for instance, built
something like an internal free market for tunneling microscopy, with multiple groups in
different parts of the building set to work on very similar tasks and compete for the
attention of senior managers. But both companies also developed an infrastructure for
STM research that allowed groups to get up to speed very quickly. For instance, Bell
Labs housed several (varying between two and four) STMs in an old tractor shed on the
edge of its property; there, microscope builders could very quickly trade ideas, materials,
blueprints, and software – very much in the same way that Quate’s students worked on
multiple microscopes at once and cannibalized parts from one project to another.31
IBM took the internal STM market/infrastructure to even greater lengths. IBM
had been first into STM, yet it took other IBM groups just as long – almost two years in
30 This analysis resonates with the early works of so-called actor-network theory: Michel Callon, “Some Elements of a Sociology of Translation: Domestication of the Scallops and the Fishermen of St. Brieuc Bay,” in Power, Action, and Belief: A New Sociology of Knowledge, ed. John Law (London: Routledge, 1986), 196-233; Bruno Latour, Science in Action: How to Follow Scientists and Engineers through Society (Cambridge, MA: Harvard University Press, 1987); and Bruno Latour, The Pasteurization of France (Cambridge, MA: Harvard University Press, 1988).31 <JG3>, <BS1>.
16
some cases – as everyone else to replicate the Zurich instrument. Thus, senior
management cast about for ways to package the tacit knowledge of instrument-building
and reduce replication time. The preferred strategy was to make semi-standardized,
batch-produced STM packages available to its researchers.32 The first was the “Blue
Box” designed by Othmar Marti, a Swiss graduate student doing doctoral work at IBM
Zurich.33 The Blue Box was primarily an electronics package – researchers constructed
the hardware themselves, often using the Zurich team’s designs. STM electronics
presented a significant challenge – complicated feedback circuitry brings the probe to the
surface, reads out and controls the tunnel current, and rasters the tip without crashing.
Later, the success of the Blue Box in allowing newcomers to work around these
difficulties inspired a more ambitious effort at IBM Yorktown. There, Joe Demuth,
manager of an STM group, assigned his postdocs to work with Yorktown’s Central
Scientific Services shop to develop and batch-produce complete STMs to “sell” to other
Yorktowners.34
By 1990, 10 to 20 of these CSS STM’s were in use at Yorktown and the nearby
Hawthorne facility; some also traveled to academic groups when postdocs left to become
professors.35 Yorktown management encouraged use of the CSS STM by making its
purchase a zero-cost budget item. Still, groups had to invest labor – usually a postdoc –
to make the microscope productive. This confronted its postdoc users with a dilemma.
They needed to creatively solve technical problems and display initiative to managers to
advance to staff positions; and advancement also required navigating competitive
32 See Philip Scranton, Endless Novelty: Specialty Production and American Industrialization, 1865-1925 (Princeton: Princeton University Press, 1997) for an analysis of batch-production.33 <OM1>, <JG1>.34 <BH1>, <RT1>, <JD2>.35 <DB1>.
17
institutional politics, where groups worked in parallel on similar projects and were
rewarded relative to each other. Postdocs using the CSS STM found they were viewed as
partisans of Demuth’s style of microscopy. To avoid alienating other factions at
Yorktown, and display their own experimental prowess, they redesigned and rebuilt large
parts of the CSS instrument.36 That is, the organization of Yorktown research kept the
CSS microscope from turning into a widely-commercialized black box.37
The CSS STM was a kind of commercialization of tunneling microscopy, for the
internal IBM market. Had Yorktown culture promoted formation of start-ups or
collaborations with instrument manufacturers, the CSS microscope could have become
the first mass-marketed STM. After the early ’90s recession made IBM leaner and more
outward-looking, Big Blue did market an AFM – Yorktown’s “SXM” – to the
semiconductor industry. This exception, though, proves the rule. The SXM was
invented by a former Quate postdoc, and owed much to Quate’s style of work. Yet
commercialization was hindered by its IBM origins – though capable of astonishing
resolution of the sidewalls of integrated circuit features, it was too finicky and unreliable
36 <JV1>, <BW2>.37 For the classic analysis of the instrument as “black box” (i.e., a technology that takes over epistemic responsibility from the experimenter by virtue of the inaccessibility of its workings), see Bruno Latour and Steve Woolgar, Laboratory Life: The Construction of Scientific Facts, 2 ed. (Princeton, NJ: Princeton University Press, 1986). My point here is that the “blackness” of the black box is continually reshaped in order to draw boundaries or form networks within an instrumental community. Commercialization can be a gradual process of making the black box less “translucent” and more less open to intervention; see Kathleen Jordan and Michael Lynch, “The Sociology of a Genetic Engineering Technique: Ritual and Rationality in the Performance of a "Plasmid Prep",” in The Right Tools for the Job: At Work in the Twentieth-Century Life Sciences, ed. Adele E. Clarke and Joan H. Fujimura (Princeton, NJ: Princeton University Press, 1992), 77-114. At the same time, the presentation of a commercial microscope architecture as “closed” (probe microscopists’ term for a black box) or “open” (their term for translucent) is a political move designed to orient that microscope toward a particular market niche/subdisciplinary audience. See Cyrus C. M. Mody, “How Probe Microscopists Became Nanotechnologists,” in Discovering the Nanoscale, ed. Davis Baird, Alfred Nordmann, and Joachim Schummer (Amsterdam: IOS Press, 2004), 119-133.
18
(it needed a Ph.D. to operate) to attract an industry devoted to tools kept in continuous
operation by relatively unskilled workers.38
In contrast, commercialization was more successful from academic STM and
AFM groups largely because of the outward-looking, multidisciplinary style they had
cultivated in order to avoid competing with the more insular and disciplined surface
scientists at IBM and Bell Labs. People like Binnig, Rohrer, Quate, and Hansma were
extraordinarily open with newcomers, freely offering them blueprints and advice in order
to build a critical mass of non-surface science probe microscopists. Thus, the circulation
of materials and ideas – a kind of “gray market” – became the norm in academic STM
and AFM.39 Software, in particular, passed from group to group, and student cohort to
cohort within research groups. Both academic and corporate groups wrote code that they
gave to collaborators, strengthening their group’s position in the instrumental community,
and ensuring access to collaborators’ modifications to the code.40 Sometimes code was
given for free, sometimes at nominal cost; profit was not the motive for dissemination.
The most well-traveled hardware innovation was the microfabricated AFM
cantilever. One perceived defect of early AFMs was that probes were laboriously hand-
made from small strips of aluminum foil with a tiny sliver of diamond glued on one side
and a tiny shard of glass on the other.41 Although these cantilevers could yield exquisite
AFM images, each required considerable time and training, and results were so particular
to one cantilever and its maker that images taken with different cantilevers were difficult
38 <KW1>, <DB2>, <JG3>.39 Davis Baird, “Scientific Instrument Making, Epistemology, and the Conflict between Gift and Commodity Economies,” Ludus Vitalis Supplement 2 (1997): 1-16.40 <MS1>.41 Diamonds were used as tips because their sharp points were less likely to wear down from repeated use than other materials. The glass on the back of the cantilever acted as a small mirror, bouncing laser light into a photodiode; the position of the reflected beam in the photodiode indicated how much the cantilever was bending (i.e., a proxy for how much the surface was pulling or pushing on the diamond tip).
19
to compare. Hand-made cantilevers sufficed early on, when every image was new and
spectacular; but as the technique matured, AFMers sought standardization. The Quate
group delivered this by integrating itself with microlithography expertise at Stanford and
around Silicon Valley. Over several years, Quate shared students with other electrical
engineering professors at Stanford, allowing them to learn AFM before going to the clean
rooms to learn to pattern and etch silicon into small, standardized batches of cantilevers.
By 1990, Quate began sending surplus probes to friends and collaborators, sometimes so
he and his students could share authorship of collaborators’ papers. Quickly, Quate-type
probes became essential to the AFM infrastructure.42
At the same time, Quate and Hansma prepared the ground for commercialization
through their practice of bringing in collaborators from a variety of disciplines. On the
one hand, these collaborators would found their own STM or AFM groups and
effectively “advertise” for the technique, building interest – what would become markets
– in probe microscopy among biologists, electrochemists, mineralogists, and so forth. On
the other hand, within the Quate and Hansma groups graduate students learned the art of
dealing with potential “customers” from other disciplines and designing microscopes
with their needs in mind. The leap from these practices to outright commercialization
was very small.
In the end, the first to make this leap was a Quate student, Doug Smith, who
founded the Tunneling Microscope Company in 1986. Notably, though, there was
wariness about this commercialization on both the supply and demand sides. Smith had
only one “employee”, a fellow student who helped put together scanners, and he recruited
customers by word of mouth. He viewed the company less as an ongoing enterprise than
42 <TA1>, <MK1>, <BD2>.
20
as a way to sweeten the hardships of graduate school – a well-circulated story is that he
sold just enough microscopes to buy a BMW before taking a postdoc. Yet despite the
small size of this venture, Quate himself was ambivalent and pushed Smith to separate
scholarship and business more cleanly – “Dr. Quate said ‘graduate students work, eat,
and sleep, and most of the time they go hungry.’ You can’t have a company and be a
graduate student at the same time, so Doug had to finish up and move out”.43
On the demand side, Smith’s customers were in much the same position as the
postdocs at IBM who were presented with the Blue Box or CSS STM. They saw a
commercial STM as a way to quickly catch up and join a hot new instrumental
community; yet they knew that if they were to join the elite of that community, they
would have to demonstrate instrument-building virtuosity, and gear their STM to
specific, arcane applications. Thus, like the batch-produced IBM instruments, Smith’s
commercial STMs were more starter kits than black-boxed devices. To use the
instrument, customers needed to construct much of it on their own.44 All Smith sold was
the microscope “head” – the piezoelectric scanner, tip, base, and vibration-isolating
stacks of Viton. Customers built the electronics themselves, customizing the microscope
for their own applications.
In debates about academic capitalism, the commodification of academic expertise
is usually treated as an undifferentiated matter, as are the consumers of that expertise.
Yet it should be clear from the STM and AFM story that the symbolic elements of the
“product” sold by members of an instrumental community is crucial and ever-changing.
Today, when customers by an off-the-shelf AFM, they generally are buying all of the
43 <MK1>.44 <JF1>, <NB1>, <RC1>.
21
expertise of microscope-building so that they won’t have to develop it themselves. Early
on that expertise was exactly what customers of commercial STMs did not want to buy;
instead, they were purchasing time and membership in a community and the opportunity
to develop their own instrument-building expertise. Consumers wanted to be on an equal
footing with producers, whether in a commercial or academic setting. Thus, there was no
“academic ethos” that was corrupted by these early ventures into batch production; yet
neither was there a “corporate ethos” demanding that firms make money from the STMs
their employees had built. Rather, in both settings it was the culture of the instrumental
community that shaped the meanings (and timetable) of commercialization.
Digital Instruments
Commercialization of the STM, then, was a process accomplished through a
series of minute, unremarkable steps – very little separated the home-built STM, made
with parts ordered from catalogs, often using blueprints given by colleagues, from the
STMs sold by Smith and (internally) IBM. The next step was only slightly more
dramatic – the founding of organizations dedicated wholly to manufacturing and selling
probe microscopes. Critics and proponents of academic capitalism both lay heavy
emphasis on the founding of start-ups; it is seen both as the best means to extract profit
from academic work, and as the ultimate distraction from the university’s pedagogical
mission. Yet both these views neglect the realities of how start-ups operate within an
instrumental community. Profit is often the least visible (and least successful) motivation
for founding a start-up; and start-ups often continue and enhance the pedagogical culture
of an instrumental community rather than despoiling it.
22
Digital Instruments, the first true start-up in the probe microscopy community,
offers vivid illustration that the most successful start-ups may be those that build outward
from an instrumental community’s gray markets and culture of circulating people, ideas,
and materials, rather than bringing a management-driven, profit-focused ideology into
that community. DI was the brainchild of Virgil Elings, one of Paul Hansma’s colleagues
in the physics department at UC Santa Barbara. Elings first contact with the STM
community, though, was Niko Garcia, a visiting Spanish academic with close ties to IBM
Zurich. After talking with Garcia and Hansma and attending the 1986 STM Conference
in Spain, Elings saw a market for an off-the-shelf STM and offered to co-found a
company with Hansma. Hansma was even more wary than Quate of commerce
encroaching on his lab’s activities, so he declined; but he gave Elings the same advice
and schematics he made available to other STMers.45 With this, Elings and his son built a
prototype in their garage and entered it in a junior high science fair (where it took last
place, since, as the judges pointed out, “everybody knows you can’t see atoms”).46
For Elings, building the prototype was a chance to make sure the Hansma design
was commercializable, but also to test – and discard – many axioms of STM-building that
accrued since 1981. Elings saw STM builders’ trade secrets as geared to instruments that
were finicky and difficult to operate; and he saw possession of these trade secrets as
limiting the STM community to those deemed “serious” enough to build their own
microscopes. Elings wanted, eventually, to make STMs specifically for non-builders
who demanded a simple-to-operate black box. Thus, he delighted in “debunking” the
STM-builders’ recipes by creating a more streamlined, easy-to-use, more durable tool.
45 <PH1>, <VE1>.46 <MT1>, <VE1>.
23
Elings wanted DI to be the first to market a commercial microscope by the annual
STM Conference in 1987. Though his plan was always to sell a computer-controlled
microscope (hence Digital Instruments), his former student (and DI co-founder), Gus
Gurley, was brought in too late to finish all the code in time. Instead, Elings marketed
the analog Nanoscope I as DI’s first product. Probe microscopists from this era – both
builders and buyers – remember their first acquaintance with the Nanoscope as a turning
point. Now, for the first time, researchers could join the STM community without having
to build any part of their microscope. Moreover, unlike Smith’s clients, DI’s customers
didn’t need to have personal ties to the community; people could (and did) simply call up
Digital Instruments and order a microscope.
Yet though it marked an important shift, the Nanoscope I still illustrates the
gradual, emergent character of commercialization. Digital retained much of the flavor of
a lab like Quate’s or Hansma’s; and Elings only slowly came to sell a more and more
complete microscope, to rely less and less on customers’ design expertise, and to view
money as the primary token of exchange. Like the CSS STM and Doug Smith’s
instrument, the Nanoscope I was more a kit than a full-fledged, black-boxed research
tool. Indeed, Elings now calls this era at DI the “toy business” – both for the Nanoscope
I’s immature design, and for its lack of “serious” applications.47 In following Hansma’s
lead, Elings designed an air STM, rather than the expensive, narrow-niche ultrahigh
vacuum instruments used at IBM and Bell Labs. This made sense in opening up a broad
market, since few disciplines were willing to deal with or pay for ultrahigh vacuum
(which, in any case, ruined samples relevant to everyone except surface scientists). Yet it
was unclear in the ’80s what air STM could be used for, or what the images it produced
47 <DC1>, <VE1>.
24
meant. Only in 1991-2 did a consensus develop that air STM was not, in fact, relying on
tunneling for its contrast mechanism, and that many well-publicized air STM images
(particularly of DNA) were erroneous. As a result, most air STMers abandoned the
technique and followed Quate and Hansma to AFM, usually by buying one of DI’s
newly-available “Multimodes” (capable of running both STM and AFM).
Thus, the Nanoscope I and other air STMs had little rigorous application, though
until 1992 it seemed possible such applications would appear if researchers could buy
cheap commercial STMs and adapt them for unforeseen purposes. The Nanoscope I was
a ‘toy’ because it was meant to be superseded, and because its buyers were envisioned as
instrument-savvy and willing to experiment playfully with their new device until
promising applications were found (not unlike DI’s engineers themselves). This
assumption of similarity between designers and users allowed new designs and
applications to flow in from the market and inform production of the more black-boxed,
all-digital Nanoscope II. Hence innovation came rapidly because participants in the
probe-microscopy field comprised an experienced and critical body. In fact, researchers
who had previously built their own STMs formed a small but elite group of DI’s early
customers.
The assumed similarities of DI’s engineers and early customers also reinforced its
self-image as a freewheeling startup that could rely on its users to make up for its lack of
marketing and customer service. Elings had no sales force – he simply advertised in
Physics Today – (“$25,000 for atomic resolution”) and orders came in. Instruments were
FedExed to buyers, who put them together and got the microscope running on their own.
Despite this minimal marketing and customer service (and limited product utility), the toy
25
business was successful. An advertisement from 1990 estimates that in the first three
years, DI sold more than 300 Nanoscopes at $25,000 to $35,000 each.48 The probe
microscopy community expanded quickly, and the center of gravity shifted as well; as
more people bought instruments, AFM and air STM began to outweigh UHV STM, and
the corporate labs became less dominant. High demand created a waiting list, prompting
a policy that researchers who wanted a microscope quickly could promise to name DI’s
founders or employees as co-authors on papers generated with Digital’s products.
DI recognized that to create a market among scientists and engineers, it had to
demonstrate its trustworthiness as a producer both of microscopes and of knowledge.49
Through its customers, DI associated credible facts with its instrument and its employees,
enticing consumers of those facts to join the probe microscopy community by becoming
consumers of its products. Notably, because the community still largely operated on a
gray market, Digital had to rely heavily on barter – such as trading waiting list position
for authorship. Such trades gradually diminished as the probe microscopy community
became more commercial; indeed, “commercialization” often signifies the narrowing of
the varieties of exchange as a technology stabilizes, rather than the encroachment of a
peculiarly corporate ethic into the academy.50
The Start-Up Era
48 From FASEB Journal, v.4, n. 13 (1990), p. 1.49 This analysis draws on Latour and Woolgar’s concept of the “cycle of credit”, but from the perspective of the instrument seller, rather than the buyer. Bruno Latour and Steve Woolgar, Laboratory Life: The Construction of Scientific Facts, 2 ed. (Princeton, NJ: Princeton University Press, 1986).50 There is a complicated relationship between “commercialization” and “stabilization” of a technology; for analyses of stabilization, see Wiebe E. Bijker, Of Bicycles, Bakelite, and Bulbs: Toward a Theory of Sociotechnical Change, ed. Wiebe E. Bijker, Trevor Pinch, and Geof Bowker, Inside Technology (Cambridge, MA: Massachusetts Institute of Technology Press, 1995) and Paul Rosen, “The Social Construction of Mountain Bikes: Technology and Postmodernity in the Cycle Industry,” Social Studies of Science 23 (1993): 479-513.
26
The end of the toy business roughly corresponded to the end of DI’s solitude in
commercial probe microscopy. By 1990, the probe microscopy community was growing
dramatically, fueled by (and fueling) the appearance of a rash of new STM and AFM
manufacturers. Their products and strategies differed considerably – some competed
with Digital Instruments for the general-purpose microscope market, some targeted
specific disciplinary niches, some made easy-to-operate black boxes, some built “open
architectures” for researchers who wanted to tinker with and modify the device. Some
survived, others floundered. All were small companies, mostly founded out of
universities specifically to make probe microscopes, though a few drifted to that product
line. No big firms made more than desultory attempts to sell STMs or AFMs – though a
few (Hitachi, IBM, Perkin Elmer) started down that road.
Crucially, these start-ups were founded for a deliciously diverse set of reasons.
Most debates about academic capitalism simply assume that, under the current set of
incentives and stimuli, professorial start-ups are an inevitability – for good or ill, the
inducements for professors to commercialize their work is simply irresistible. Maybe so,
but this assumption looks less reliable when the contingent and often counter-intuitive
reasons why people found start-ups are examined. Digital Instruments provides a striking
example. DI and Elings thrived at UCSB’s disreputable margins. When he arrived in the
late ’60s as a brash, confrontational professor, it was hoped Elings would build UCSB’s
reputation in high energy physics. His swagger, though, led to conflict with his
department, which sidelined him into running their less prestigious Master’s of Scientific
Instrumentation program – a lucrative but unloved backwater.51 Uncowed, Elings
51 <JW1>.
27
transformed the master’s program into his personal empire and a fountainhead for patents
and start-up companies.
In the master’s program students from many educational backgrounds (biologists,
engineers, even psychology majors) learned to build all kinds of measurement
technologies – not just research instruments but also industrially-relevant meters and
tools. Initially, Elings relied on orthodox classroom instruction; but soon, he drifted
toward an alternative method that prized tacit over formal knowledge, participation over
instruction. Instead of textbooks and lectures, he simply connected students with
professors on campus who needed instruments built and let them learn by doing.
Because student projects were based on finding solutions to real problems faced by local
researchers, they often yielded technologies Elings could market to those researchers’
subdisciplines. Students learned how to understand customers’ needs and design
technologies to answer them. This made former master’s students the most important
source of early employees for all Elings’ ventures, especially Digital Instruments.
So UC Santa Barbara did, in a way, encourage creation of DI, though no school
would replicate their path. By sidelining a brilliant but difficult professor to the poorly-
regarded master’s program, they encouraged him to reject campus culture, denigrate
academically-instilled formal knowledge, and be receptive to the commercial possibilities
of the tacit knowledge his students accrued. Moreover, in making clear Elings’
commercial ventures hindered his academic career, the UCSB physicists made it more
likely his next enterprise – Digital Instruments – would be his bridge to leaving
academia. Tension between Elings and UCSB even smoothed technology transfer from
Hansma to DI, since Elings’ hostility toward academic researchers meant he rejected
28
Hansma’s designs until they had been engineered to look more like commercial products
than most home-built instruments.
Disgruntlement of a different kind also fueled both DI and some of its
competitors. Several of the early STM and AFM manufacturers were founded in the
heart of the West Coast military-industrial complex. Graduate students who had grown
accustomed to the picturesque surroundings and lifestyle of southern California often
sought employment nearby – usually with defense firms like Lockheed and Hughes. Yet
defense work galled many of these engineers, driving them back to probe microscopy.
Much of Elings’ early work-force came to DI for this reason; and in Los Angeles Paul
West, one of John Baldeschwieler’s former postdocs at Caltech, grew so frustrated with
defense work that he started his own probe microscopy company, Quanscan.
In several cases, start-ups positioned themselves relative to each other in ways
that mirrored the relationships between the academic groups with which they were
associated. Digital Instruments may not have been officially affiliated with the Hansma
group at first (indeed, even at the best of times, there was always some suspicion between
the two groups), and Elings was certainly proud of the ways the Nanoscope differed from
Hansma’s microscopes; but DI’s products bore an obvious genealogical kinship with
Hansma’s STMs and AFMs, and DI drew on Hansma’s reputation in the community. In
return, Hansma’s design innovations spread much farther and faster than those of
professors not affiliated with a start-up. Thus, Hansma’s peers in the probe microscopy
elite were more enthusiastic than he about helping former personnel found start-ups. For
instance, two Stanford postdocs, Sung-Il Park and Sang-Il Park (no relations) started Park
Scientific Instruments in 1989 with Quate’s help and quickly became the major employer
29
of Quate group veterans; PSI’s designs, even more than DI’s, were ported straight from
the academic group, and the research in new applications conducted there often picked up
where Quate’s own research left off.52 Commercialization is the politics of academic
research by other means.
Similarly, just as Baldeschwieler’s groups always lagged behind Quate’s and
Hansma’s in popularizing its discoveries and innovations, the company he helped Paul
West found, Quanscan, lagged behind DI and Park Scientific in marketing
commercialized versions of the Caltech designs. As Bourdieu might put it, relations
defined by intellectual capital in the academic field mapped onto relations of commercial
capital in the start-up field.53 These relations could sometimes be based on collaboration
as much as competition. For instance, Stuart Lindsay, a physics professor at Arizona
State University, had been an early collaborator of Hansma’s, one of the first in the long
series of visitors to UCSB who helped adapt the STM to new applications and
subdisciplines – in Lindsay’s case, for electrochemistry and for biological materials.
Once Hansma’s designed were commercialized by DI, Lindsay pressed Elings to adapt
the Nanoscope for Lindsay’s colleagues in electrochemistry and biophysics – to no avail,
since Elings was usually hostile to adapting the Nanoscope for anyone (DI had a strict
no-custom-instruments policy), especially when the suggestion came from outside the
company. So Lindsay founded his own company, Molecular Imaging, to make
attachments to the Nanoscope that would make it more compatible with electrochemistry
and biophysics – attachments which DI grudgingly distributed for a few years until it
developed its own competing line.54
52 <DB3>, <FG1>, <JN1>.53 Pierre Bourdieu, “The Forms of Capital,” in John Richardson, ed., Handbook of Theory and Research for the Sociology of Education (New York: Greenwood Press, 1986), 241-258.54 <SL1>.
30
Lindsay’s other motivations for founding a start-up say a great deal about how
commercialization and pedagogy fit together in instrumental communities. Long before
Molecular Imaging, his group – like Hansma’s at UCSB – had become a center for
distributing blueprints and (especially) software to new STM builders. One of Lindsay’s
technicians, Uwe Knipping, developed one of the first and most sophisticated computer-
controlled microscopes; Knipping’s software formed the basis for Lindsay’s academic
network-building, but it also caught the eye of two local entrepreneurs, Larry and Darryl
McCormick, who founded a company, Angstrom Technology, to commercialize it.55
As it turned out, Knipping’s architecture was far too sophisticated for a
commercial instrument, and the enterprise failed. But Lindsay had gotten a taste for how
network-building in the academic domain might be enhanced by commercialization. So a
few years later when he happened on a former postdoc who was having trouble finding
work, Lindsay decided to built a new company – Molecular Imaging – around him as an
extension of the work going on at ASU.56 This is probably the classic – if woefully
understudied – story of commercialization: a professor’s technicians and graduate
students make a widget, then the professor’s colleagues call up asking for their own
widget (or blueprints thereof), a student start making batches of widgets in their garage,
and eventually – whether to help position the professor within that instrumental
community or to give lab personnel needed work – the widget-making is spun off as its
own kind of organization. Commercialization here makes the academic research more
streamlined and focused by being an extra bin in which to throw specific kinds of much-
needed people and community-building work.
55 <SL1>, <JA1>.56 <SL1>, <TJ1>.
31
In only a few cases were probe microscope start-ups inspired by the incentives to
commercialize and culture of entrepreneurialism so central to the debate about academic
capitalism. Paul West, for one, explicitly saw entrepreneurship as a challenge and path to
personal growth, and the Baldeschwieler group had a long history of spinning off
commercial enterprises.57 Similarly (but to a much lesser extent) the Parks and their
employees molded PSI in the image of Stanford’s tradition of commercialization, and
Quate’s particular tradition of sending former students to the most powerful firms in
Silicon Valley. Yet these start-ups’ success was seemingly in inverse proportion to the
entrepreneurial intent of their founding. Quanscan (and its successor company,
Topometrix) was always the most business-like – had the most venture capital money, the
most MBAs, the slickest advertising, etc. But these proved a drag on the company’s
fortunes – the bankers continually interfered in operations, the MBAs had trouble
understanding the values of an instrumental community that they had never participated
in, and the advertising alienated many potential customers.
Park Scientific, like DI, was a more rough-hewn affair – Sung-Il Park’s barber
was hired as the office manager, for instance, and the senior management were Quate
students with no business training.58 Indeed, the Parks carved a niche for their
instruments by cultivating an image of themselves as interested much more in technically
sweet innovations than in mundane moneymaking. As “gentleman scientists,” they could
speak as peers with other researchers and capture customers’ trust. Park Scientific gained
a reputation for making “builders’” instruments – well-crafted, reliable, with enough
idiosyncracies and innards showing to be reminiscent of a microscope made by a
57 <PW2>, <JB1>, <GA1>.58 <DB1>, <BP1>.
32
graduate student. Park was even willing to work with individual customers to build a
microscope for a specific application (something DI never did) – if the engineering
required a certain finesse. Yet ultimately this sapped the company. Park rode to some
success on the same hostility to management expertise as DI; but Digital ultimately won
out because (once the toy business was over) it displayed an almost equal hostility to the
expertise of its customers. Where Park was willing to relive the days of the Quate lab by
respecting the knowledge of foreign disciplines, DI made one type of microscope for
everyone, and hid the workings of that instrument completely from customers’ view.
This attitude allowed DI to break into the industrial market, where it could sell radically
more, and more expensive, microscopes to companies that usually wanted low-level
technicians to learn how to use the instrument in a day or two – market conditions for
which Park was wholly unprepared.
As much as DI eventually prospered by distancing itself from Hansma’s academic
model, though, the company’s success hinged equally on continual borrowings of culture,
people, and inventions between start-up and academic lab. By chance, these borrowings
were facilitated by Elings and Hansma’s convergence on similar pedagogical
philosophies. Both me saw tacit, rather than formal, knowledge as primary in instrument-
building – Elings because of his work in the instrumentation master’s program, Hansma
because the contours of the STM community had pushed him to encourage undisciplined
instrument-building rather than disciplined instrument-use. This shared emphasis on the
tacit meant both men took in people with diverse and unusual educational backgrounds:
junior high students, river guides, undergraduates, yoga instructors, retirees, psychology
majors, and historians.59 This diversity was unthinkable at other centers of probe
59 <HH1>, <JM3>, <DB2>, <MT1>, <BD2>, <PH1>.
33
microscopy. Orienting to tacit knowledge also meant both DI and the Hansma group
thrived on self-cultivating activities seemingly unrelated to technical matters; Hansma’s
group found technical inspiration in pastimes such as woodworking, meditation,
photography, yoga, river-rafting, etc; while DI held weekly “inventing sessions” where
employees brainstormed solutions to esoteric (i.e. non-AFM-relevant) technical questions
(e.g. “how do you make a self-balancing laundry machine”) to become better inventors
and hone their skills at weathering Elings’ intense skepticism.60
As members of DI and the Hansma group became aware of parallels between
their organizational styles, they appropriated these similarities to accelerate the two-way
flow of people, materials, designs, and knowledge. After the initial phase (when most DI
employees were Elings’ former master’s students), several Hansma graduates, postdocs,
and collaborators took high-ranking jobs at Digital. Individuals on both sides
collaborated to transform Hansma’s research into commercial products; for instance, the
Hansma AFM (on which DI’s fortunes eventually rested) was turned into a product
through negotiations between Barney Drake (Hansma’s technician) and James Massie (a
former Elings student) over which elements of the Hansma design were indispensable
and which were too finicky for anyone but the graduate students who built them.61 As
DI’s sales increased, the Hansma group kept its place at the forefront of the AFM
community through its steady supply of DI instruments and the ability of Hansma’s
students and postdocs to go up the road to DI to scavenge parts and advice.62 That is,
whatever his initial reservations about commercialization, Hansma came to see the
60 <JH1>, <DB2>, <PM1>.61 <BD2>, <JM3>.62 <JH1>.
34
partnership with DI as a way to position himself – intellectually and socially – within his
instrumental community.
In turn, once the “toy business” ended in the early ‘90s, Elings began to imitate
Hansma’s tactic of bringing in postdocs to guide instrument-builders’ efforts. DI built its
own group of researchers from biophysics, magnetics, and polymer chemistry, who (like
Hansma’s postdocs) worked with instrument-builders, developed and published on new
STM and AFM applications, and traveled to give talks and attend conferences to spread
word about the technique.63 Though DI was a profit-making venture, its success arose
partly from the Hansma group practices that it mirrored – doing research, publishing
articles, training and “graduating” employees. These practices were then widely
emulated by the other start-ups.
This kind of cultural/organizational parallelism from university to start-up is well-
documented in other fields of commercialization, especially biotechnology; there, it was
used to entice professors to exit the ivory tower, leading to trouble when professor-
entrepreneurs were reluctant to turn their companies from research to profit-making.
With STM and AFM, in contrast, corporate-academic isomorphism was successful less as
a deliberate strategy than as a contingent and emergent harmonization of practices. It
was a largely accidental outcome of the organization of this instrumental community that
Quate, Hansma, and other academics promoted a gray market of circulating people,
practices, and technologies that fostered successful commercialization.
Conclusion
So what does probe microscopy tell us about commercialization of academic
knowledge and the value of corporate-academic linkages? First, the development of
63 <SM2>, <MA1>.
35
probe microscopy shows how thoroughly – yet intricately and indirectly – the corporate
and academic worlds are connected. The locus of ‘academic research’ is much wider
than the university campus, just as the locus of ‘commerce’ is wider than the for-profit
business. Instrumental communities and other informal organizations are distributed
across academic and corporate institutions. Commercialization – the transformation of
academic research into commerce – is not a simple pipeline from university to firm.
Commercialization can play many roles within an instrumental community, and academic
research can be traded for many things other than money. Attempts, therefore, to directly
stimulate and accelerate the transformation of academic research into cash may well
backfire. As we have seen, it was the looser, indirect ties between corporate and
academic groups that fostered the growth of STM and AFM and encouraged startups to
emerge from universities, rather than direct pressure from corporations or overt
incentives from governments and universities.
Thus, proponents of academic entrepreneurialism should be wary of focusing too
narrowly on increased profit as the fruit of a commercialized university. As we’ve seen,
trading goes on all the time in instrumental communities; the token of exchange is usually
a mix of knowledge, prestige, personnel, time, materials, money, opportunity, etc. The
popularity of various forms of barter changes as the instrumental community changes;
commercialization can restrict some exchanges and make money-based trades more
prevalent. Few instrumental communities reach this point, though. Even within the
probe microscopy community, only the atomic force microscope and the magnetic force
microscope have been commercial successes; the STM, which provided the first product
for microscope manufacturers, was effective in training engineers to build microscopes,
36
but never found industrial application. The presence of gray markets in instrumentation
can enhance national economic growth over time; yet university administrators who hope
that this or that gray market can be converted into a profit-making start-up to enhance
local, short-term economic growth will almost always be disappointed.
Moreover, development of an entrepreneurial instrumental community may
require that its members be drafted from less profitable fields where commercialization
did not occur. The STM and AFM community, for instance, initially drew on its
members’ expertise in low-energy electron diffraction, sandwich tunnel junction
spectroscopy, and field ion microscopy – instrumental communities with poor records of
commercialization; later, STM and AFM pulled in participants from many fields (surface
science, biophysics, mineralogy, electrochemistry, polymer science – some more
commercialized than others) who aided groups like Quate’s and Hansma’s in their gray
market activities. Instrumental communities in which the cultural map is unconducive to
profit-making nevertheless provide the infrastructure and knowledge/labor pool for
communities in which profit may be enormous. Policy-makers should not think they can
predict which will be which; nor are they likely to succeed if they encourage only the one
at the expense of the other. Policy makers may be best advised to encourage professors
to foster gray markets within their instrumental communities – whether as consumers,
producers, or both. Gray market activities of trading research materials, people, and
components of technologies enlarge the outlook of academic research. By focusing on
the wider instrumental community surrounding a technology, we can see that the
university may actually be more influential in maintaining a pool of skeptical,
37
independent consumers who can threaten start-ups with the prospect of making their own
tools or even founding their own firms.
Finally, both opponents and supporters of corporate involvement in university life
have seized on grains of truth. Supporters have it right that corporate-academic linkages
are desirable, even necessary, for research and innovation. There was no golden age
when faculty operated independent of firms, pursuing disinterested research; knowledge
production in physics, engineering, and chemistry was always aided by faculty consulting
and trading of personnel and ideas. The oft-criticized commercialism of the “biotech
revolution” merely extended long-standing entrepreneurial practices into molecular
biology. The STM and AFM case does, however, give reason for opposing the notion
that universities should be run as businesses, squeezing profit where they can and
operating along the “rational” lines of modern management. The probe microscopy
community developed rapidly because participants could point to different institutional
poles – corporations, universities, national labs. At times, innovation occurred because
these poles were opposed – as when Hansma and Quate shifted from surface science and
UHV STM to new designs and applications. At other times, innovation occurred because
participants strung out hybrid forms between these poles – the gray market of software
trading, the CSS STM, and the “toy business”. Instrumental communities rely on a
variety of actors, contained in different kinds of institutions. If all these institutions are
run on the same highly-managed, profit-driven model, then the movement of people and
ideas, and the production of new technologies, will likely be hindered.
Appendix
38
Interviewees listed by alphanumeric, name, positions held over the period covered by the
interview, and date of the interview. All interviews conducted by the author.
AG1: Andy Gewirth: Hansma collaborator; University of Illinois; 6/26/01BD2: Barney Drake: Hansma group technician; UCSB; 10/18/01BH1: Bob Hamers: Yorktown researcher; University of Wisconsin; 5/9/01BP1: Becky Pinto: Stanford; Park Scientific; KLA-Tencor; 2/3/04BS1: Brian Swartzentruber: Bell Labs technician; University of Wisconsin; Sandia
National Laboratory; 1/10/03BW2: Bob Wolkow: IBM Yorktown; Bell Labs; NRC Canada; 5/22/01CG1: Christoph Gerber: IBM Zurich technician; 11/12/01CP1: Craig Prater: Hansma graduate student; Digital Instruments engineer; 3/19/01DB1: Dawn Bonnell: Yorktown postdoc; University of Pennsylvania; 2/26/01DB2: Dan Bocek: UCSB undergraduate; DI engineer; Asylum Research; 3/23/01DB3: David Braunstein: Stanford; Park Scientific; IBM San Jose; 4/3/01DC1: Don Chernoff: Sohio Research; Advanced Surface Microscopy; 9/5/01DF1: Dave Farrell: Burleigh Instruments; 5/29/01DR1: Dan Rugar: Quate student; Almaden researcher; 3/14/01FG1: Franz Giessibl: IBM Munich; Park Scientific; Uni Augsburg; 11/16/01GA1: Gary Aden: Topometrix executive; 3/12/01HG1: Hermann Gaub: Ludwig-Maximilians Universität; 11/14/01HH1: Helen Hansma: UCSB professor; 3/19/01JA1: John Alexander: Angstrom Technology; Park Scientific; KLA-Tencor; 10/15/01JB1: John Baldeschwieler: Caltech; 3/28/01JC2: Jason Cleveland: Hansma student; Digital Instruments; Asylum Research; 3/20/01JD2: Joe Demuth: Yorktown manager; 2/22/01JF1: John Foster: Quate student; Almaden researcher; 10/19/01JG1: Jim Gimzewski: IBM Zurich researcher; UCLA; 10/22/01JG3: Joe Griffith: Bell Labs; 2/28/01JH1: Jan Hoh: Hansma postdoc; Johns Hopkins; 6/10/02JM1: John Mamin: UC Berkeley; IBM Almaden; 3/15/01JM3: James Massie: Elings master’s student; DI engineer; 10/18/01JN1: Jun Nogami: Quate postdoc; Michigan State; 6/28/01JV1: John Villarrubia: Yorktown postdoc; National Institute of Standards and
Technology; 6/28/00JW1: Jerome Wiedmann: Elings master’s student; DI employee; 10/18/01KW1: Kumar Wickramasinghe: Stanford; IBM Yorktown; 2/23/01MA1: Mike Allen: UC Davis; Digital Instruments; Biometrology; 10/12/01MK1: Mike Kirk: Quate student; Park Scientific Instruments; KLA-Tencor; 10/12/01MS1: Miquel Salmeron: Lawrence Berkeley National Laboratory; 3/9/01MT1: Matt Thompson: Digital Instruments; 2/26/01NB1: Nancy Burnham: Naval Research Lab postdoc; Worcester Polytechnic; 2/20/01OM1: Othmar Marti: IBM Zurich student; Hansma postdoc; University of Ulm; 11/16/01
39
PH1: Paul Hansma: UC Santa Barbara; 3/19/01PM1: Pete Maivald: DI employee; 10/18/01PW2: Paul West: Caltech; Quanscan; Topometrix; Thermomicroscopes; 3/30/01RC1: Rich Colton: Naval Research Lab; Baldeschwieler collaborator; 6/27/02RT1: Ruud Tromp: Yorktown researcher; 2/23/01SG1: Scot Gould: Hansma student; DI employee; Claremont McKenna; 3/27/01SL1: Stuart Lindsay: Hansma collaborator; Arizona State; Molecular Imaging; 1/6/03SM2: Sergei Magonov: Digital Instruments; 3/21/01TA1: Tom Albrecht: Quate student; Almaden researcher; 3/14/01TB1: Thomas Berghaus: Uni Bochum; Omicron; 11/19/01TJ1: Tianwei Jing: Arizona State; Molecular Imaging; 1/7/03VE1: Virgil Elings: UC Santa Barbara; Digital Instruments; 3/20/01
40